In a network system using any network, access to a server is generated from terminals distributed and located at many remote sites. In a case where performance evaluation of servers connected to the terminals distributed and located at many remote sites through a LAN is executed, each of the terminals inputs thereinto a program for performance evaluation of a server which the terminal accesses, runs the program for the performance evaluation of the servers, and collects a result of the response from the server. In this case, it is required for an operator to take the program to each of the terminals and to bring back the result.
In order to solve the problem described above, a method in which an evaluation command is sent to each of the terminals through a network and a result of monitoring is collected is considered. As for the network system based on the conventional technology, for instance, a network OS evaluation system on a local area network was disclosed in Japanese Patent Laid Open Publication No. 177893/1994. Detailed description is made hereinafter for the conventional technology with reference to the related drawings.
FIG. 22 is a block diagram showing configuration of the conventional type of network system. In FIG. 22, designated at the reference numerals 101a to 101m are a plurality of target servers as objects for monitoring, and at 102 a test controller for registering thereto items for testing and evaluating the target servers 101a to 101m as objects for monitoring, generating test commands according to the registered test items, and transmitting the commands to test servers 103a to 103n. The reference numerals 103a to 103n indicate a plurality of test servers for testing a network in relation to the specified target servers 101a to 101m according to the test commands transmitted from the test controller 102, and the reference numeral 104 indicates a test monitor for displaying and recording a result of the test or situations of the test for the target servers 101a to 101m transmitted from the test servers 103a to 103n. The reference numeral 105 indicates a LAN for connecting the target servers 101a to 101m, test controller 102, test servers 103a to 103n, and test monitor 104 to each other.
In the conventional type of network system, a network OS as an object for monitoring is installed in the target servers 101a to 101m. At first, the test controller 102 generates test commands according to items for the test previously registered in the target servers 101a to 101m, and the generated test commands are transmitted to the test servers 103a to 103n.
Then, when the test servers 103a to 103n receive the test commands transmitted from the test controller 102, a specified network test is executed to the specified target servers 101a to 101m according to the received test commands. Also the test servers 103a to 103n periodically transmit test situations executed or a result of the tests for the target servers 101a to 101m, to which network tests were executed, to the test monitor 104.
The test monitor 104, when having received the test situations or the result of the tests executed to the target servers 101a to 101m transmitted from the plurality of test servers 103a to 103n, displays and records the test situations and the result of the tests executed to the target servers 101a to 101m.
In the conventional type of network system, one unit of test controller 102 can uniformly control tests for the target servers 101a to 101m operated by each of the test servers 103a to 103n, so that there is an advantage therein that an operator for monitoring is not required to operate for each of clients respectively.
In the conventional type of network system described above, the object is to monitor and evaluate a network OS as objects for monitoring installed in the target servers 101a to 101m, so that transfer of test commands from the test controller 102 to the test servers 103a to 103n, access from the test servers 103a to 103n to the target servers 101a to 101m by executing tests, and reporting of a result of execution from the test servers 103a to 103n to the test monitor 104 are concurrently executed.
In the conventional type of network system described above, the target servers 101a to 101m, test controller 102, test servers 103a to 103n, and test monitor 104 are directly connected to a LAN 105, so that a transfer rate in the network is sufficiently high. For this reason, even if test command transfer from the test controller 102 to the test servers 103a to 103n, access from the test servers 103a to 103n to the target servers 101a to 101m by executing tests, and reporting of a result of execution from the test servers 103a to 103n to the test monitor 104 are concurrently executed, any problems described later hardly occur.
However, in a case where the target servers 101a to 101m, test controller 102, test servers 103a to 103n, and test monitor 104 are not directly connected only to LAN 105, but are linked to each other by using a wide area network, the data transfer rate in the wide area network is generally low. And for this reason, when transfer of test commands from the test controller 102 to the test servers 103a to 103n, access from the test servers 103a to 103n to the target servers 101a to 101m by executing tests, and reporting of a result of execution from the test servers 103a to 103n to the test monitor 104 are concurrently executed, the data transfer rate becomes lower than that in a case where data is simply transferred from the test servers 103a to 103n to the target servers 101a to 101m, due to an increased data volume being concurrently transferred. Accordingly, in a case where a period of time is measured a period of time from a point of time when access from the test servers 103a to 103n to the target servers 101a to 101m is executed until a point of time when a response comes back therefrom, the data transfer rate during the period of time is measured as lower than the actual rate because also different data flows therethrough concurrently, and for this reason a period of time required to transfer the necessary data can not correctly be obtained.
As described above, since a plurality types of different data flow concurrently, a load applied to the test server as well as to the test controller 102 increases, and for this reason performance of the target servers 101a to 101m can not correctly be monitored.